R/RFM.R

In astrosyam/RFM:

#' RFM
#' @description  RFM (Recency, Frequency, Monetary ) analysis
#' @param data Dataset for analysis.
#' Should contain at least three columns : customer name, date of purchase and amount
#' @param analysisDate Date used of recency calculations. Defaults to current date.
#' @param customer Name of column containing name of customer. Default : customer
#' @param date Name of column containing date of transaction. Default : date
#' @param revenue Name of column containing the amount of transaction. Default : amount
#' @keywords RFM
#' @export
#' @examples
#' rfm(sales)
#' @author Syam Murali

rfm = function(data, analysisDate = Sys.Date(), customer ="customer",
               date = "date", revenue  ="amount")
{

  if(ncol(data) != 3){
    stop("Data should have customer, amount and date columns")
  }

  data = data[,c(customer, date, revenue)]
  names(data) = c("customer", "date", "revenue")

  if(length(data$customer) == 0 | length(data$date) == 0 | length(data$revenue) == 0){
    stop("customer, amount or date column not specified")
  }

  recency = function(data){
    aggregate(date ~ customer, data , min)
  }

  frequency = function(data){
    aggregate(date ~ customer, data[,1:2] , length)
  }

  monetary = function(data){
    aggregate(revenue ~ customer, data[,c(1,3)] , sum)
  }

  score = function(scoreData){

    maxQuartile = length(unique(scoreData))

    if (maxQuartile > 5){
      maxQuartile = 5
    }
    lowerLimit = (5 - maxQuartile) + 1
    scoreLabel = as.character(cut(scoreData,
                                  breaks = c(quantile(unique(scoreData),probs= seq(0,1,length.out = maxQuartile + 1))),
                                  labels = seq(lowerLimit,5),
                                  include.lowest = TRUE))
    as.numeric(scoreLabel)
  }


  # Formatting data fields
  data$customer = as.factor(data$customer)
  data$date = as.Date(data$date, '%d/%m/%Y')
  data$revenue = as.numeric(data$revenue)

  # Set Analysis Date
  analysisDate = as.Date(analysisDate)
  data$date = as.numeric(difftime(analysisDate,  data$date, units = "days"))

  # Calculating Recency, Frequency and Monetary values for each customer
  Recency   = recency(data)
  Frequency = frequency(data)[,2]
  Monetary  = monetary(data)[,2]

  rfm_data = cbind(Recency, Frequency, Monetary)
  names(rfm_data)[2] = "Recency"

  RFM_Score = do.call(cbind, lapply(rfm_data[,-1], function(x) score(x)))
  RFM_Score[,1] = 6 - RFM_Score[,1]
  RFM_Score = as.numeric(apply(RFM_Score,1,paste, collapse = ''))

  data = cbind(rfm_data, RFM_Score)


  # Sorting data frame in decreasing order of RFM value
  data = data[order(data$RFM_Score, decreasing = TRUE),]

  return(data)
}